ssh-agent, protocol 2, openssh-2.3.0p1

[Markus Friedl]

you have to try a recent snapshot if you need agent-fwding
in ssh2: http://bass.directhit.com/openssh_snap/

On Thu, Dec 28, 2000 at 11:12:36AM -0700, Gary Shea wrote:
> Hi --
> 
> I saw an email on December 26th in the openssh-unix-dev mailing list
> archived on MARC, indicating that agent forwarding is indeed not working
> for 2.2, but that it is working for 2.3.  That email referred to a user
> with 2.3 clients and 2.2 server.  I am running the 2.3 client AND server
> and am having a similar problem.  The only unusual aspect of my installation
> is that I'm using port 24 until I convince myself that this is all something
> I'm doing wrong, and switch to openssh!  Apologies in advance if I'm
> offbase...
> 
> I'm a moderately experienced ssh user, been running various versions for
> about three years, and have been using agent forwarding for the last
> couple of years with ssh.com's ssh1 and ssh2.  This is my first dalliance
> with openssh.
> 
> There are two machines, A and B.  A is running openssh-2.3.0p1 compiled
> by me on Mandrake Linux 7.2 (RedHat 7?).  B is also running
> openssh-2.3.0p1 compiled by me on Solaris 5.8 (an UltraSparc box).
> On A and B, the /etc/ssh/ssh_config and .ssh/ssh_config
> have 'Host *' followed by 'ForwardAgent yes' entries, and no
> other Host lines.  As far as I can see, there are no sshd_config
> directives that affect agent forwarding.
> 
> The Mandrake installation (on A) uses ssh-agent to set up KDE,
> so I see an ssh-agent running and have a corresponding SSH_AGENT_PID
> and SSH_AUTH_SOCKET in the shell environment.  I ssh-add
> .ssh/gary_shea.dsa (my identity file, set up with IdentityFile2 in
> .ssh/ssh_config).  ssh-add -l shows the identity.  If I then 'ssh -A'
> to machine B, I do not need to type the passphrase, indicating that the
> ssh-agent works to that extent.  BUT, my environment on B
> does not contain the SSH_AGENT_PID and SSH_AUTH_SOCK entries.
> 
> I am forwarding X11 also, and the X11 forwarding appears to be working,
> based on the presence of DISPLAY in the environment on B, and the
> characteristic delay starting vim on B.
> 
> Is there something I'm missing?
> 
> 	Gary
> 
>